Thorn in the Side: The 2003 Zürich-Oerlikon (Switzerland) Train Collision
Oerlikon is a district of 23214 people (as of 2018) in northern Zürich in the north of Switzerland, located 17km/10.5mi southwest of Winterthur and 25km/15.5mi south of the German border near Jestetten (both measurements in linear distance).
Zürich-Oerlikon station is a large eight-track (at the time of the accident six-track) passenger station, one of the oldest and most important stations in Switzerland (opening in 1855). Sitting a short distance north of Zürich main station is a “choke point” in the city’s railway system, with essentially all north- and northeast-bound trains leaving Zürich main station (a terminus station at ground level with pass-through tracks below) having to pass through it before leaving the city.
The trains involved
Travelling northbound from Zürich main station to Konstanz was Regio Express “Cityvogel” (“City Bird”) 1629, consisting of nine four-axle passenger cars. The first car on the train was a second class Swiss “Einheitswagen 1” (shortened EW I), a standardized four-axle passenger car introduced in the late 1950s. The cars weight 30 metric tons at 23.7m/78ft long. They can reach a top speed of 140kph/87mph, enough for most passenger services even decades after their introduction. The car at the front of the Cityvogel at the time of the accident still wore its original all-green livery with dark yellow doors.
Pulling the northbound train at the time of the accident was SBB (Swiss national railway) series 420 number 11113. The Series Re 420, originally called the Re 4/4 II, is a four-axle multipurpose electric locomotive introduced in 1967. It’s the backbone of the Swiss railway network, pulling everything from freight trains and regional services to long distance express trains. The locomotives measure 14.9m/49ft in length at a weight of 80 metric tons and can reach a top speed of 140kph/87mph thanks to a power-output of 3700kW/4962hp. The locomotive involved in the accident was one of the earliest ones, recognizable by only having a single pantograph on the roof. The entire train had a listed weight of 365 metric tons according to the report.
Coming the other way was Express train 2583 from Schaffhausen to Zürich main station, pulled by SBB series 460 number 003. The series Re 460 is a four-axle electric locomotive for passenger trains measuring 18.5m/61ft at a weight of 84 metric tons, a power-output of 6100kW/8294hp allows them to reach as much as 200kph/124mph even with a heavy passenger train in tow. Being introduced in 1991 the Re 460 took over most higher-class passenger services, limiting the Re 420 to regional and less prestigious passenger services and the freight-sector.
The southbound express train consisted of 13 four-axle passenger cars. Important for this accident are cars 6 and 7, which were second class four-axle express passenger car type Bpm 51. These cars were created in the mid-1980s by converting former compartment cars to open space interiors. The cars, originally introduced in the late 1970s, seat 77 passengers at an overall length of 26.4m/87ft and weight 43 metric tons each. The total weight of the express train is listed at 565 metric tons.
On the 24th of October 2003 at approximately 4:50pm SBB Re 420 11113 is entering track E9 at the shunting facilities outside Zürich main station, pulling up to two waiting passenger cars. The driver shuts down the locomotive to allow a shunting worker safe coupling of the two waiting cars to the locomotive while he moves to the other cab, at the same time cars 3–9, coming from Geneva as train 2727, are added to the far side of the two waiting passenger cars, completing the train. The shunting-worker who coupled the locomotive to the two waiting cars also checks the correct coupling of the second car to the incoming train cars, connecting the train cars as well as the pneumatic lines and electronic connections. After starting up the locomotive a test of the braking-system yields sufficient results on the third attempt. Unbeknownst to anyone involved none of the attempts were done correctly, as the locomotive delivering the train cars 3–9 was still hooked up to that train at the time of the test. Furthermore, the shunting worker had neglected to properly control the connection between the two waiting cars. Had he done this he would have realized that the pneumatic valve between the two cars was closed, during the brake-test the locomotive from train 2727 actually fed the pneumatic lines in cars 2–9, not locomotive #11113 as it should be.
The train driver, possibly distracted by correcting a harmless error the conductor had made when filling out paperwork, apparently didn’t notice that the pneumatic system refilled too fast between tests. At 5:20pm the locomotive from 2727 is removed from what is now the back of the train and, with a shunting worker keeping watch from car 9, the completed “Cityvogel” was pushed into track A3 at the main station at low speed, stopping just fine and allowing passengers to board the train. At the station a new driver takes over the locomotive, with the delivering crew reporting that the train is ready to depart but that the braking power is “not ideal”. However, with the measurements being still within tolerances the new driver has no reason to delay departure with further tests, following protocol he can depart and climb the incline out of the main station before testing the different braking-systems once he reaches Zürich-Oerlikon station, the first scheduled stop. After all, had the brake-system not performed within tolerances the train wouldn’t have been pushed into the station in the first place.
At 5:37pm the Cityvogel departs Zürich main station and accelerates to 50kph/31mph. Soon the train passes Wipkingen station (approximately 2.8km/1.75mi after departing) where the driver accelerates to 70kph/43.5mph as the train dives into a tunnel. Inside the tunnel the driver is supposed to reduce the speed by 10kph/6.2mph, due to the incline simply reducing the throttle-input achieves this deceleration. Intending to use the stop at Oerlikon station for another test the driver first initiates a normal stop, dropping air pressure by 0.7bar. Finding a near-nonexistent deceleration on the now level track he soon reduces air pressure by another 0.3bar. This would be a rather rapid deceleration any other day. Still finding insufficient braking-power the driver triggers an emergency stop, dumping the entire air pressure. At that point he realizes that something had gone very wrong, braking-power wasn’t “not ideal”, it was abysmal. He triggers the “shunting brake”, an additional independent pneumatic brake on the locomotive, right as he enters Oerlikon station. Braking power is still violently insufficient, and the driver realizes that from this point on he is essentially just a passenger. The locomotive has all brakes fully engaged and was just being pushed along by the heavy train cars. As the southbound train comes into view the Cityvogel’s driver sounds the locomotive’s whistle in a last-ditch effort to warn the other driver, who had already realized that the oncoming Cityvogel was too fast and getting too close. He had subsequently also triggered an emergency stop, hoping the oncoming Cityvogel would stop just short of his train or miss it entirely. He almost gets his wish, with the two locomotives narrowly missing each other.
At 5:44pm the driver of the southbound express train feels what he later calls a “slight jolt” as his locomotive enters Oerlikon station. Halfway down his train Re 420 11113 has crashed into the side of car 6 at 31kph/19mph, crushing the entrance-area of the train car and ripping it off the cars ahead. After the initial impact the locomotive moves back out of the oncoming train car before striking it a second time and becoming lodged in the rear section of the destroyed passenger car. The stop is sudden enough for the leading passenger car of its own train to climb the frame of the locomotive and crush the rear driver’s cab. A Czech tourist seated near the initial point of impact is killed instantly while 23 people are injured, 17 of which severely. Throughout the night another 22 survivors seek medical attention on their own. In total around 1400 people had been on the two trains. The force of the collision sets off the fire extinguisher in the locomotive’s cab, the injured train driver has to wait for rescue covered in the powdery substance.
So close to the station (the southbound train stopped at the platform) evacuation of most passengers was a swift undertaking. Meanwhile the driver retrieved the paper printout from the data-logger in his cab, checked the transformer for a possible oil leak and then waited for the arriving responders to help him leave the destroyed locomotive. The driver had been on the left side of the cab, leaving him with sufficient survival room while the collision with the side of the oncoming train and the impact of its own train into the back crushed both the right hand side of the leading cab as well as the entire rear driver’s cab.
Within the hour all passengers have been removed from the train with the survivors being evaluated and released or transported to surrounding hospitals. While the southbound train’s sixth car had had little to offer against the heavy and stiff locomotive, especially its frame as it tore through the side of the car, the Cityvogel’s cars had remained largely intact, retaining sufficient survival space and easing evacuation of the passengers. The leading end of the southbound train’s sixth car had suffered a catastrophic structural failure, with the locomotive “taking a bite” out of the corner of the car, destroying the floor, walls and roof in the entrance-area and the first few rows of seats before being pushed back out of the train car, swaying to the other side and striking the train a second time a few meters further down the side of the car.
So called “Flankenfahrten”, side impacts at a sharp angle, are an engineering nightmare for train cars as it is incredibly difficult to offer any side impact protection on what is essentially a large rectangular metal box. The damage that led to the death of the passenger can’t be blamed on poor or outdated engineering, the impact of the locomotive simply meant too high a force pushing into the side of the train car on too small an area for the frame of the train car and the motion of the train derailing to absorb it. A straight impact (head on or rear-ending) is much easier to engineer protection against, and while on the EW 1 this was mostly down to the stiff frame modern train cars also feature designated crumple zones at either end to absorb forces before they can be inflicted on the passenger area.
By the time daylight arrives the trains have almost entirely been removed from the site, 11113 is the last vehicle to be towed away. It’s fairly obvious that brake-failure caused the collision, but the question is how that failure happened. Paperwork and statements by those involved say that the train passed the brake-tests and was supposedly assembled correctly. An early suspicion is that someone sabotaged the train, closed the brake-valves after the brake-system had been tested and signed off on. This is disproven when no unauthorized people can be placed at the site of the brake-test and none of those involved had a motive to cause a train wreck. Lastly, taking the suspicion that train 2727’s locomotive had still been attached to the train into account, investigators figure out that a brake-test can’t have been conducted had both locomotives fed the entire train with pressurized air. Looking into the documentation of where the locomotive went after disconnecting from the train the investigation creates a timetable and eventually proves that with near-certainty the shunting crew had accidentally forgotten to disconnect the locomotive from the assembled train before testing the brakes. Adding to that a simple human error of only checking the valves between the locomotive and the leading car as well as between the two groups of cars this meant that the brake-test could be passed despite the valves between cars 1 and 2 being closed the entire time. As it was the investigation placed blame on the worker who should have checked the brake-valves between cars 1 and 2, not just between the locomotive and the first car as well as between the two groups of cars.
It took until April 2008 for a trial to start, at which point the then 56 years old SBB-employee was put on trial for negligent manslaughter in one case and several cases of negligent cause of bodily injury. The public prosecutor’s office expressed that the trial dealt with an extreme case of lazy negligence, and demanded a sentence of 1 year in jail, set out to probation. The defense disagreed with both the public prosecutor and the report, demanding the defendant to be relieved of criminal guilt. Furthermore the defense criticized the SBB’s supposedly overly tight schedule, along with a ban of post-departure brake tests at the station. Had the driver been allowed to test the brakes right after departure (still at the platform) the train would have, at worst, stalled and stopped at the incline out of the station. It would’ve never gotten to Oerlikon-station, would’ve never picked up nearly as much speed.
Zürich’s regional court surprisingly sided with the defense, relieving the defendant of criminal guilt. There was no proof that he had closed the valves, and it was deemed possible that, due to routine, he was sure that he had checked the valves and they had been open. Who had closed them couldn’t be determined. The public prosecutor immediately announced to appeal the sentencing, only for the decision to be confirmed in a retrial 2 months later. While the employee was partially responsible, he wasn’t at criminal fault for the ensuing events. As such, no one ever faced legal consequences for accident or the death and injury it caused. The SBB did change their procedures after the accident, independently from the trial, by 2007 drivers were now meant to check the brakes right after departure. The new guideline came 4 years too late for the 22 years old Czech tourist.
The locomotive and leading car from the Cityvogel were scrapped after the investigation concluded, along with cars 5–6 from the southbound train. Total damage to material was listed in the report at 4.8 million Swiss Francs (4.39 million Euros/5.15 million USD).
The SBB retired their last EW 1 passenger car in spring 2021, apart from a handful being preserved by different owners in a historic capacity most of what remained of 1208 cars made is headed to the scrapyard after over 60 years in service. Bpm 51 have similarly started to disappear from Swiss rails, with services being largely taken over by bilevel trains or Mark 4 Einheitswagen. Since 2010 the SBB is gradually retiring the Re 420, while most get scrapped some actually get sold publicly, with prices starting at 450 thousand Swiss Francs. Only the Re 460 is still in full service, with the SBB’s fleet currently undergoing a refurbishment program giving them more efficient electronics and a new livery. Locomotives more or less closely based on the Re 460 are also in service in Norway, Finland and HongKong.